The goal of this project is providing machines with the ability of understanding what a person is experiencing from her frame of reference, taking into account the scene context: where is this person, what is this person doing, how does this person look, etc.
You can find more information about this project on this website.